• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.05a
• /
• pp.1461-1464
• /
• 2004

본 논문에서는 LAN 상에서 유휴 PC들을 사용하여 PC Cluster를 구성한 후 이를 사용한 내용기반의 화상 정보 검색(CBIR) 시스템에 관한 연구를 수행하였다. LAN 상의 유휴 PC 들은 인터넷상의 연산 자원들보다 안정되고 신뢰성이 있기 때문에 복잡한 보안 기법을 사용하지 않아도 되며 또한 연산시간이 유휴시간으로 고정되어 있기 때문에 네트워크의 부하 및 노드의 부하를 고려하는 복잡한 부하 균등화 기법이나 스케쥴링 기법이 필요로 하지 않는 특징을 갖는다. 내용기반의 화상 정보 검색은 화상 데이터의 대용량 특징으로 인해 화상 특징 추출 및 유사도 계산을 위해 많은 연산을 필요로 한다. 특히 다양한 내용기반의 정보 검색 서비스를 지원하기 위해 다중 특징(Multiple Features)을 동시에 추출하고자 할 때 연산시간은 급격히 증가한다. 따라서 이러한 내용 기반의 화상 정보 검색 시스템을 구현하기 위해 저비용의 고성능의 PC Cluster를 사용하여 전체 연산시간을 단축하고 실시간 정보검색이 가능하도록 하는 연구를 수행 하였다.

• Journal of information and communication convergence engineering
• /
• v.20 no.2
• /
• pp.96-102
• /
• 2022

We aim to build predictive models for Airbnb's prices using a GPU-accelerated RAPIDS in a big data cluster. The Airbnb Listings datasets are used for the predictive analysis. Several machine-learning algorithms have been adopted to build models that predict the price of Airbnb listings. We compare the results of traditional and big data approaches to machine learning for price prediction and discuss the performance of the models. We built big data models using Databricks Spark Cluster, a distributed parallel computing system. Furthermore, we implemented models using multiple GPUs using RAPIDS in the spark cluster. The model was developed using the XGBoost algorithm, whereas other models were developed using traditional central processing unit (CPU)-based algorithms. This study compared all models in terms of accuracy metrics and computing time. We observed that the XGBoost model with RAPIDS using GPUs had the highest accuracy and computing time.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.8
• /
• pp.2042-2060
• /
• 2013

In wireless sensor networks (WSNs), energy efficiency is one of the most essential design considerations, since sensor nodes are resource constrained. Group communication can reduce WSNs communication overhead by sending a message to multiple nodes in one packet. In this paper, in order to simultaneously resolve the transmission security and scalability in WSNs group communications, we propose a hierarchical cluster-based secure and scalable group key management scheme, called HRKT, based on logic key tree and route key tree structure. The HRKT scheme divides the group key into cluster head key and cluster key. The cluster head generates a route key tree according to the route topology of the cluster. This hierarchical key structure facilitates local secure communications taking advantage of the fact that the nodes at a contiguous place usually communicate with each other more frequently. In HRKT scheme, the key updates are confined in a cluster, so the cost of the key updates is reduced efficiently, especially in the case of massive membership changes. The security analysis shows that the HRKT scheme meets the requirements of group communication. In addition, performance simulation results also demonstrate its efficiency in terms of low storage and flexibility when membership changes massively.

• IEIE Transactions on Smart Processing and Computing
• /
• v.6 no.3
• /
• pp.200-209
• /
• 2017

In this paper, we investigate the tradeoff between security and reliability for a multi-hop protocol in cluster-based underlay cognitive radio networks. In the proposed protocol, a secondary source communicates with a secondary destination via the multi-hop relay method in the presence of a secondary eavesdropper. To enhance system performance under the joint impact of interference constraint required by multiple primary users and hardware impairments, the best relay node is selected at each hop to relay the source data to the destination. Moreover, the destination is equipped with multiple antennas and employs a selection combining (SC) technique to combine the received data. We derive closed-form expressions of the intercept probability (IP) for the eavesdropping links and the outage probability (OP) for the data links over a Rayleigh fading channel. Finally, the correction of our derivations is verified by Monte-Carlo simulations.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.6 no.3
• /
• pp.103-109
• /
• 2010

Total performance is improved by minimizing the channel interference between links in wireless mesh networks (WMNs). The paper refines on the CB-CA [1] to be suitable for multi-radio multi-channel (MRMC) WMNs. The CB-CA is the cluster-based channel assignment algorithm for one radio three channel WMN based on IEEE 802.11b/g. The CB-CA does not perform the channel scanning and the channel switching between the cluster heads (CHs) and the edge gateway nodes (EGs). However, the use of co-channel for links between CHs and EGs brings the problem of channel interference among many nodes. We propose and evaluate an improved CB-CA algorithm to solve this problem in MRMC WMNs. The proposed algorithm discriminates between transmission channel and receive channel and assigns channels to each interface randomly and advertises this information to neighbor clusters in order to be assigned no-interference channel between clusters. Therefore, the proposed algorithm can minimize the interference between clusters and also improve QoS, since it can use multiple interfaces and multiple channels.

• Journal of the Optical Society of Korea
• /
• v.20 no.2
• /
• pp.276-282
• /
• 2016

A 2 Gbit/s visible-light communication (VLC) scheme using time-frequency color-clustered (TFCC) multiple-input multiple-output (MIMO) based on blue, cyan, yellow, and red (BCYR) light-emitting diodes (LEDs) is presented. In the proposed scheme, BCYR LEDs are employed to form four different color clusters. Data transmission using the four color clusters is performed in MIMO, so that the scheme achieves a very high speed of data transmission. Moreover, the scheme employs the TFCC strategy to yield high performance in terms of bit error rate (BER). TFCC operates in such a way that the original data and the two delayed versions of the data are multiplied by orthogonal frequencies and then transmitted using a specific color of the BCYR LED. In the receiver, color filters are employed to detect the data transmitted from the desired cluster. Selection combining (SC) is also performed to yield a diversity effect within each color cluster, to further improve the performance. Performance evaluation demonstrates that the proposed TFCC MIMO VLC offers a data rate of 2 Gbit/s and a bit error rate of 4×10^-5, at an E_b/N_o value of merely 3 dB.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.9
• /
• pp.1658-1666
• /
• 2009

This paper proposes an approach of Mult_HPSO based on the PC cluster system to reduce or remove the stagnation on an early convergence effect of PSO, reduce an execution time and improve a search ability on an optimal solution. Hybrid PSO(HPSO) is combines the PSO(Particle Swarm Optimization) with the mutation of conventional GA(Genetic Algorithm). The conventional PSO has operated a search process in a single swarm. However, Multi_PSO operates a search process through multiple swarms, which increments diversity of expected solutions and reduces the execution time. Multiple Swarms are composed of unsynchronized PC clusters. We apply to SCED(security constrained economic dispatch) problem, a nonlinear optimization problem, which considers line flow constraints and N-1 line contingency constraints. To consider N-1 line contingency in power system, we have chosen critical line contingency through a process of Screening and Selection based on PI(performace Index). We have applied to IEEE 118 bus system for verifying a usefulness of the proposed approaches.

• Journal of the Korean Academic Society of Industrial Cluster
• /
• v.1 no.1
• /
• pp.29-46
• /
• 2007

The purpose of this paper is to analyses current status and issues of cluster formation and extract policy implications. To this end, the questionnaire which surveyed the level of cluster formation were executed targeting the actors of regional innovation projects(RIPs). The results show that the situations and development stage of the cluster formation between capital region and non-capital region, large cities and small and medium sized cites are different. The level of clustering is also satisfactory, which is a requirement for cluster formation at its early stage. However, the capacity for phase II of cluster growth is not sufficient yet in terns of relationships between ventures and large corporations, institutions supporting management, finance and marketing, researchers from each individual sector of strategic industries and spin-off fines. Therefore, RIPs should be promoted with different policy tools for various regions that are devised according to the varying development stage of each region. The location of RIPs should be determined considering efficiency rather than equity, clustering rather than decentralization, and specialization rather than multiple development. In the long term, developed regions should pursue balanced regional development, with underdeveloped regions targeting specialization.

• Preventive Nutrition and Food Science
• /
• v.7 no.2
• /
• pp.201-206
• /
• 2002

Findings from European and American studies have indicated such health behaviors as smoking, drinking, lack of exercise, and insufficient intake of grains, fruits and vegetables as risk factors for hypertension. However, because dietary pattern and health behaviors of Korean differ from people of other countries, the risk factors for Koreans could be different. Therefore, the aim of this study was to identify food consumption patterns and health behavior characteristics of Korean hypertensive adults. Data on food consumptions and other health behaviors such as smoking, drinking, and exercise together with development of hypertension were collected from 597 persons aged 20 to 69 years in a local community. The first stage of analysis utilized cluster analysis to aggregate individuals into different health behavior and food consumption groups. Four health behavior groups were characterized by passive cluster, smoker cluster, fitness cluster and drinker cluster. Food consumption patterns of the subjects were also aggregated into 4 different clusters: dairy cluster, grain & vegetable cluster, fruit cluster, and fish & meat cluster. Then univariate analysis was followed to identify the variables associated with hypertension. The final stage of analysis was the identification of the relative importance of the variables selected from the univariate analysis on hypertension, using multiple logistic analysis. The results showed that heavy drinking was the most significant health behavior associated hypertension, which was similar to the findings in European and American studies. However, unlike the findings from foreign studies, grain and vegetable consumption appeared to be a risk factor for hypertension. A possible reason for the contradictory results between Korean and Western studies may be the dependence of Koreans on white rice as the major staple food, and/or the frequent consumption of salted vegetables, rather than fresh vegetables as is customary in Europe and America.

• Culinary science and hospitality research
• /
• v.17 no.2
• /
• pp.74-84
• /
• 2011

The purpose of this study is to understand the relationship between educational environment services offered and student satisfaction with them. A survey was conducted to 314 students in a cooking and food service program. To achieve the purpose of this study, SPSS 17.0, a statistics program, was used for a descriptive analysis, a frequency analysis, a factor analysis, a multiple regression analysis, a cluster analysis and a One-Way ANOVA analysis. The hypothesis test result of the effect of educational environment services on satisfaction showed that there was a partially significant effect in satisfaction with facilities and lecturers(p<0.01) as well as administrative services (p<0.01). In the cluster analysis of educational environment services, cluster 1 was classified as a mistry cluster, cluster 2 as a satisfied cluster, cluster 3 as a dissatisfied cluster and cluster 4 as a positive cluster. They showed a statistically meaningful result in the one-way ANOV A(p<0.01).